JPH0722208B2 - Semiconductor sensor device - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to an improvement of a semiconductor sensor device such as a magnetic sensor, a temperature sensor, a pressure sensor, and a strain sensor, which has a capability of detecting a variation amount in an external environment.

<Prior Art> Various types of semiconductor sensor devices that output converted electric quantities such as voltage and current corresponding to changes in the amount of fluctuation given from the external environment such as magnetic field, temperature, pressure or strain. However, these days, especially, it is suitable with the aid of fine processing technology such as photolithography so that it is very small and can be integrated with other circuit systems on the same substrate as needed. There is one that is created on a semiconductor substrate.

As a typical example of such a magnetic sensor device using the Hall effect, its main structure is generally as shown in FIG.

The figure is a plan view, and the paper surface of the drawing may be considered to be the surface of the semiconductor substrate or a plane parallel to this, but here, as it is usually called a Hall element, it is called a semiconductor magnetic sensor device. However, only a single magnetic sensor cell is shown.

Actually, a plurality of cell structures of the same kind are integrated on one semiconductor substrate, or further, by arranging a plurality of them in an appropriate direction and performing an appropriate electrical and circuit treatment,
There is a device that can detect not only the magnitude of the applied magnetic field but also the direction of the applied magnetic field. However, in each case, the point of the present invention is the same. Think

First, the Hall element 10 has a current path 20 formed in a surface parallel to the surface of the semiconductor substrate, and generally this is the same as it is, and the substantial variation detection area (that is, the magnetic field in this case). Detection area) 40. However, in some cases, the magnetic field detection region 40 may be intentionally formed in the middle of the current passage 20 by a separate material. However, even in such a case, in this document, the current path has a magnetic field detection region, and is therefore collectively described.

Bias current is applied to the current path 20 or the magnetic field detection area 40.
Since I _S is applied and used, terminal regions 23 and 24 are formed at both ends in the lengthwise direction thereof, and electrodes 25 and 26 made of a metal or other suitable conductive material are provided on the surfaces thereof, respectively.

The magnetic field detection region 40 or the current passage 20 in the case shown is of course made of a material capable of exhibiting the Hall effect, but in order to take out the converted electric quantity as a result of the magnetic field detection, that is, the Hall voltage, it is in the middle of its length. A pair of signal extraction line-shaped regions 31 and 32 are drawn out at a right angle to the current path 20 in a plane parallel to the substrate.
Electrodes 35 and 36 made of an appropriate material are formed on the surface regions of the terminal regions 33 and 34 of 1, 32, respectively.

Generally, in such a sensor device, the pair of signal detection terminals (electrodes) 35 and 36 are often configured as a so-called balanced type. That is, even when the bias voltage V _S or the bias current I _S is applied to the main current path 20, a DC potential difference does not occur between the pair of signal detection terminals 35 and 36 when the magnetic field is not detected. As planned.

For example, in the case of the Hall element shown in the figure, if a DC voltage may appear between the pair of signal detection terminals 35 and 36 even when a magnetic field is not applied, it is necessary to detect alternating magnetic fields. In any case, the detection of the DC magnetic field is actually impossible. As is well known, such a bias voltage is called an offset voltage and is disliked.

Therefore, in order to satisfy this, as a structural factor, the pair of signal extraction regions 31 and 32 is geometrically connected to the main current passage 20 or the magnetic field detection region 40 at the same longitudinal position. . That is, the illustrated relative position error ΔL is set to zero as much as possible.

Now, assuming that this requirement is satisfied, the voltage V _S is applied between both electrodes 25 and 26 of the current path 20 in such a structure, and the bias current I _S is caused to flow, whereby the magnitude of the magnetic field applied from the outside is increased. It can be detected by converting it into Hall voltage.

That is, when a magnetic field to be detected is applied to the detection area 40, the current is biased in the current passage 20 or the magnetic field detection area 40 due to the well-known Hall effect, and accordingly the signal extraction area 3 is generated.
Since a corresponding Hall voltage V _H can be obtained between both electrodes 35, 36 via 1, 32, the magnetic field B to be detected is this Hall voltage.
It can be calculated from V _H and the following equation 1).

Where B is the desired magnetic field (Gauss), d is the thickness of the current path 20 (cm), R _H is the Hall constant, and R _H = 1 / en (cm ³ / coulomb). (1) <Problems to be solved by the invention> In the Hall element as described above, it has already been mentioned that the relative connection positional relationship between the pair of signal extraction regions 31 and 32 with respect to the current path 20 is an important issue. did. The position deviation amount ΔL when viewed in the length direction of the current passage 20 should ideally be zero.

However, in reality, with such a semiconductor sensor device having a conventional configuration, the positional deviation amount ΔL cannot be made to be zero, and even it is difficult to make it sufficiently small.

Therefore, conversely, let us consider how much such a positional deviation ΔL actually affects the detection characteristics of the element.

Both signal extraction regions 31, 32 when the magnetic field B is not applied
The bias voltage (offset voltage) V _O generated between is obtained by the following equation 2), where L is the effective length of the current path 20.

However, in the case of the Hall element structure described above, all of its main parts include a planar structural factor, and therefore, as a result of being actually applied by applying ordinary photolithography, the geometrical dimension deviation ΔL is 1 μm at maximum. It also becomes a degree.

On the other hand, since the length L of the current passage 20 is usually set to about 100 μm, if the power supply voltage V _{S is set} to 10 V, for example, the above equation 2)
Therefore, the bias voltage V _O is about 100 mV.

At this time, if the semiconductor of the current path or the magnetic field detection region is silicon, the Hall voltage for a magnetic field of 1KG is on the order of 10 mV, and in such a case, the unbalanced voltage V _O becomes 10 times the signal voltage. . This is, of course, beyond the permissible range and in particular the DC magnetic field is virtually undetectable.

Such a problem may similarly occur in other semiconductor sensor devices having a fluctuation amount detecting portion formed in a plane parallel to the surface of the semiconductor substrate, such as a semiconductor temperature sensor device, a semiconductor pressure or strain sensor device, and the like. It's a problem.

For example, in the temperature sensor device and the pressure sensor device as described above, generally, four resistance elements are assembled in a bridge,
Physical or mechanical follow-up processing or special mechanical means is added to one of them as a target fluctuation amount detection section, and the signal extraction line areas are each one from the equilibrium point of this bridge,
It is constructed so that a total of one pair can be pulled out.

In such a case, the connection position of each signal extraction area to the corresponding current passage portion also needs to be selected in a position where there is no geometrical relative displacement, but in reality, existing planar processing is required. As long as it relies on photolithography as the microfabrication technology according to (1), its positional accuracy is limited, and the unbalanced voltage or bias voltage V _O cannot be suppressed sufficiently.

Expanding this discussion, it goes beyond the problem of balanced types. Even if there is only one signal extraction area extracted from the fluctuation amount detection area, for example, in a sensor device of the type that reads the voltage between the ground and the like, the geometric extraction position of the signal extraction area is The characteristics can be sufficiently influenced by the deviation from the design value, and similarly, if the conventional planar microfabrication technology, that is, photolithography must be relied on, the defect in accuracy cannot be wiped out.

The present invention has been made in view of these points, and in a semiconductor sensor device having a fluctuation amount detection section in the middle of a current path, the accuracy of the connection position of the signal extraction region to the current path is determined by a low-precision method such as photolithography. It is another object of the present invention to provide a semiconductor sensor device which can be defined by the impurity double diffusion technique and the thin film deposition technique which can exhibit much higher accuracy even at the present time.

<Means for Solving Problems> In order to achieve the above object, the present invention proposes a semiconductor sensor device having the following configuration.

A current passage (20) inside the semiconductor substrate (11) and extending in the thickness direction of the semiconductor substrate (11); the current passage (29) itself, or the current passage (2)
A variation amount detection region (40) provided in the middle of (0) for detecting a variation amount applied from the external environment; provided at both ends of the current passage (20) and biased to the variation amount detection region (40). Terminals for passing current (25,26)
And; electrically connected to the fluctuation amount detection region (40) inside the semiconductor substrate (11) and extending from the connection portion in a direction orthogonal to the current passage (20) to a signal detection terminal region. (33,34), and the above fluctuation amount detection area (4
0) from the fluctuation amount detection area (40) to the signal detection terminal area (3).
A semiconductor sensor device comprising: a signal extraction region (31, 32) for transmitting to a signal (3, 34).

<Operations and Effects> The above-mentioned variation amount referred to in this document, starting from the gist of the present invention, is the amount of each dimension such as a magnetic field, temperature, pressure or strain to be externally detected, as is already clear. , The converted quantity of electricity corresponding to the amount of fluctuation refers to voltage, current, etc., and the semiconductor substrate refers to one in which the entire substrate is made of a single material, one in which a plurality of layers are laminated, or one on a heterogeneous substrate. It is a general term including those having a thin film formed on.

However, in any case, in the present invention, the current passage is formed so as to extend inside the thickness direction of the semiconductor substrate, and the signal extraction region is located in the semiconductor substrate inside the semiconductor substrate in the predetermined thickness direction. The fact that the connection is made at the height position means that it is not necessary to use photolithography, which is a planar processing technique, for setting the connection position, and the impurity double diffusion technique, which is far more precise than photolithography, can be used. This means that a film processing or film forming method in the thickness direction, such as a thin film deposition technique, can be applied, and thus sufficient positional accuracy can be obtained.

In practice, when the present invention is applied to the above-described Hall element or the like, the relative positional deviation ΔL between the signal extraction regions is easily 10
It can be around 0Å.

Also, since the effective length of the current path is in the thickness direction (depth direction) of the semiconductor substrate, it can be made sufficiently short,
For this reason, the same amount of current as in the past can be applied to the current path,
The applied voltage can be one to two orders of magnitude smaller than the conventional one.

As a result, as a result of reducing both the effective length of the current path and the applied voltage, even if the ratio V _S / L in the above formula 1) is considered to be constant, the positional deviation ΔL is further reduced in the present invention. Since the offset voltage V _O can be extremely reduced,
It can be greatly reduced to about 1/100. Therefore, as described above with respect to the Hall element, when compared under the same conditions that the applied magnetic field is 1 KG and the Hall voltage is 10 mV, the Hall element according to the present invention is effectively less sensitive than the conventional Hall element. It is 100 times higher.

It is clear that such a remarkable effect can be generally developed. The semiconductor sensor device to which the present invention should be applied has the basic effect of the present invention that the connection position of the signal extraction region with respect to the current path having the fluctuation amount detection region can be set with much higher accuracy than before. It is a universal effect that does not depend on the type of operation or on the principle of operation.

<Embodiment> FIG. 1 shows a Hall element 10 as a single body as an example of a semiconductor sensor device according to the present invention.

In addition, regarding the reference numerals used for the respective constituents in the drawing, the same reference numerals are given to those having the same or similar functions as those of the respective constituents in the above-mentioned conventional example. Of course, their physical, mechanical, and geometric configurations often differ as a result of the application of the invention.

The current passage 20 having the magnetic field detection region 40, or in the case of this embodiment, the magnetic field detection region 40 as it is is the substrate.
It is formed so as to extend from the surface of 11 to the inside in the thickness direction of the substrate, and is connected to the terminal region 24 of the same conductivity type, but preferably of low resistance, formed on the back surface side of the substrate.

Similarly, the terminal region 23 provided on the other end side of the current passage 20
Also, it is preferably formed on the surface of the substrate 11 as a region having the same conductivity type as the current path and low resistance.

The pair of terminal areas 23, 24 are provided with electrodes or terminals 25, 26 for connection to an external circuit, respectively. The terminals 26 for the terminal area 24 are usually formed on the back surface of the substrate 11 by a suitable method. It can be constituted by a metal or other conductive material surface 26.

The current path 20 in such a structure can be configured by existing techniques such as impurity diffusion from the semiconductor substrate surface, ion implantation, and selective epitaxial growth, and the terminal regions 23 and 24 are similarly well-known and suitable methods. Can be formed by.

In addition, the portion where the low resistance region 24 is formed is the first semiconductor substrate.
The current path 20 can also be formed by selectively adopting the above method in the reverse conductivity type thin film selected as 24 and grown thereon. In the figure, as an example, the current path 20 is n-type, therefore the low resistance terminal regions 23, 24 are n ⁺ type, and the conductivity type of the peripheral portion of the current path 20 is p-type, which is the opposite conductivity type. The case is shown.

The signal extraction regions 31 and 32 are connected to the current path 20 extending in the vertical direction inside the semiconductor substrate 11.

Such a structure starts from the surface of the semiconductor substrate 11 through the current path
An impurity of the same conductivity type as that of 20 is diffused through an opening (diffusion window) having a predetermined area so as to reach a predetermined depth d1, and then an impurity that defines the opposite conductivity type is diffused through the same diffusion window as above from the depth d1. By double-diffusing over a shallow predetermined depth d2,
Signal extraction area 3 with area thickness dt determined by dt = d1-d2
1, 32 are formed inside the semiconductor substrate at a predetermined height position (a predetermined depth position when viewed from the semiconductor substrate surface).

Thereafter, signal detection terminal regions 33 and 34 for establishing conduction to an external circuit are formed in the signal extraction regions 31 and 32 by an impurity diffusion technique from the substrate surface, etc., and the exposed surface is made of a suitable conductive material. By attaching the electrodes 35 and 36, a Hall element capable of performing the same function as the conventional Hall element already described in terms of an equivalent circuit is completed. Of course, this signal extraction area 31, 32
The low resistance regions 37 and 38 may be formed in the terminal regions 33 and 34 related to.

As described above, the hall element 10 configured according to the present invention is the most different from the conventional element, as is apparent from the above-described configuration, the connection position of the pair of signal extraction regions 31, 32 to the current path 20. However, it can be defined in the height direction inside the semiconductor substrate 11 by the impurity double diffusion technique.

Therefore, the displacement ΔL is the diffusion depth of each diffusion region.
Since it falls within the error range in this kind of double diffusion technique for d1 and d2, which is extremely small,
In practice, the positional deviation ΔL can be easily contained within about 100Å.

In addition, such accuracy can be obtained in the same order of magnitude not only by the above-mentioned double diffusion but also by other existing thickness direction thin film processing methods or forming methods such as ion implantation and epitaxial growth. Therefore, in other words,
The structure itself of the present invention includes such high precision.

Further, as described above, in the present invention, the effective length of the current path 20 is taken in the thickness direction of the semiconductor substrate 11, so that it can be sufficiently shortened and reduced to the order of several μm. be able to.

In fact, in the prototype of the applicant, the effective length of the current passage 20 could be 5 μm. If the length of the current path 20 can be reduced in this way, the bias current I _S
It is also possible to reduce the applied voltage V _S between the terminals 25 and 26, which are provided at both ends of the current path 20 and which are required to pass the current, and reduce the load on the peripheral bias circuit, thus reducing the circuit system. The effect that can be simplified is also produced.

When the applied voltage V _S = 5V was applied to the prototype device with an effective current path length of 5 μm and the magnetic field was detected in the direction parallel to the surface of the semiconductor substrate and orthogonal to the drawing sheet, the Hall voltage V _H could get 100 mV / 1KG, while unbalanced voltage V _O could be suppressed to 10 mV.

This substantially corresponds to the fact that the sensitivity can be increased 100 times under the same conditions as compared with the conventional Hall element.

By the way, in such a Hall element, as described above, it is desirable that not only the applied voltage can be lowered but also the applied current can be reduced. This is because the efficiency of energy utilization can be improved and the peripheral circuit system can be similarly simplified.
For that purpose, the thickness of the current passage may be reduced. In the illustrated case, the thickness of the current passage means a cross-sectional area in a direction parallel to the lying direction of the semiconductor substrate.

Therefore, as another preferred embodiment of the present invention, an embodiment as shown in FIG. 2 can be provided.

The figure (A) is a plan view, and the figure (B) is a main part structure unified by the same conductivity type for understanding, especially the magnetic field detection region.
1 is a perspective view and a schematic view of a current passage 20 which also serves as 40, a pair of signal extraction regions 31, 32, and respective terminal regions 23, 24, 33, 34 of the current passage and the signal extraction region.

As can be seen from these figures, in this embodiment, after forming the constituent volume of the current passage 20 in the first embodiment by a suitable technique as described with respect to the embodiment shown in FIG. , The impurity of the opposite conductivity type is diffused by double diffusion of impurities from the same diffusion window, and the core structure 21 of the opposite conductivity type is formed inside, so that the current path 20 is thick around the core. By making the wall thin, the effective thickness is reduced. The thickness is second
It is shown by the symbol W in FIG.

By doing so, the bias current value required to obtain the same detection sensitivity can be sufficiently reduced.

In the embodiment shown in FIG. 2, the components by other reference numerals not described are the functionally and structurally first components.
For the components with the same reference numerals in the embodiment shown in the figure,
You can think that they are almost the same or the same.

FIG. 3 shows a structure suitable for constructing the present invention as a balanced pressure sensor or temperature sensor.

As shown in the equivalent circuit of FIG. 1B, in a semiconductor sensor device that generally detects such temperature or pressure, four resistance element parts are assembled in a bridge, and two of these four are connected in series. A voltage V _S is applied to both ends of each branch, and a bias current I _{S is made} to flow.

Two resistance element parts 20a, 20b; 20c, 20 in each branch
One end of each of the signal extraction regions 31 and 32 is connected to a connection node between d, and a corresponding voltage as a converted quantity of electricity is obtained between the terminals 33 and 34 on the other end side.

Of course, one of the four resistance element units, for example, the resistance element unit 20d, is mechanically or geometrically configured to directly or indirectly contact the external environment and to detect a target variation amount such as temperature or pressure. Must be structured.

Applying the present invention, such an equivalent circuit can be obtained as a structure as shown in FIG.

As before, the same reference numerals are used for the corresponding components as in the previous embodiment. However, of course, in this embodiment, the variation amount to be detected is not the magnetic field as described above, and therefore the variation amount detection area 40 provided in the current passage 20 is provided.
Respectively become a pressure sensitive region and a temperature sensitive region depending on the case.

Inside the semiconductor substrate 11, a pair of current passages 20, 20 are formed so as to extend in the thickness direction and are appropriately laterally separated from each other.

The current path 20 is made of an appropriate material so as to be equivalent to the series configuration of the resistance element portions 20a, 20b; 20c, 20d having pressure or temperature sensitivity, but the physical configuration method is not As described in the above embodiments, the impurity diffusion technique, the ion implantation technique, the epitaxial growth technique and the like can be selectively adopted.

Similarly, the signal extraction regions 31 and 32 connected in the lengthwise direction of the current passage 20 have a depth d1 from the surface of the semiconductor substrate 11.
The position and thickness dt can be set accurately by the double diffusion technique with different d2. Generally, in such a bridge, the connection positions of the signal extraction regions 31 and 32 are set in both current passages 20 and 20 so that the resistance element ratios of the resistance element portions 20a, 20b; 20c, 20d in each branch in the steady state are equal. Selected at the same depth position.

Terminal regions 33, 34 connected to the respective signal extraction regions 31, 32 and exposed on the surface of the semiconductor substrate, terminals or electrodes 35, 36 formed thereon, and terminal regions 23, 24 at both ends of the current path, the same electrodes. Similarly, 35 and 36 can be formed by the method already described.

The characteristic additional configuration for the temperature or pressure (strain) detection device is that the resistance element portion 20d of one branch is
The semiconductor substrate portion on one side in the horizontal direction (the right hand side in the figure) is cut open by, for example, digging open, and the side surface is formed of a suitable insulating film 50.
Facing the external environment through, and the side surface that is indirectly but facing the external environment, the side of the facing side is similarly processed by suitable etching means or the like, and dug down, That is, the vacuum space 51 is formed. The layer 52 that seals over the vacuum space 51 and protects the top of each terminal is composed of a suitable heat insulator or vacuum seal.

In such a sensor device 10, the temperature or pressure fluctuation acting on the side surface facing the external environment causes the resistance element portion as the fluctuation amount detection region 40 in the one current passage 20.
The change in the resistance of 20d can be read in the form of a change in the bridge balance voltage, that is, a potential difference between the terminals 33 and 34.

Of course, the mechanical configuration in the illustrated case is an example, and the mechanical configuration itself for causing an external variation amount to act on a specific resistance element portion is not directly specified by the present invention. The present invention is characterized by the fact that the current path having the fluctuation amount detection region is configured to extend inside the semiconductor substrate in the thickness direction.

In any of the first to third illustrated embodiments, if the sensor cells shown in the figure are arranged in a suitable plane, for example, at least two sensor cells are arranged so that the directions for receiving the detected fluctuation amount are orthogonal to each other, the fluctuation amount can be increased. It is possible to provide a semiconductor sensor device that identifies not only the size of the above, but also its application direction within the plane.

On the contrary, it is also possible to provide a semiconductor sensor device as a single body by forming a plurality of integrated formations on a semiconductor substrate and then cutting out one by one. Of course, it is of course possible to mount the sensor device of the present invention on the same substrate as other circuit systems, and in addition, any of the above-described embodiments was a semiconductor sensor device configured as a balanced type. If the sensor device is such that the connection position of the region with respect to the current path must be strictly defined, good results can be obtained by applying the present invention even in the unbalanced type.

[Brief description of drawings]

FIG. 1 is a schematic configuration diagram of the semiconductor sensor device when a magnetic sensor is configured as an embodiment of the present invention, and FIG.
As another embodiment of the present invention, a schematic configuration diagram of an embodiment in which the thickness of a current passage is reduced, and FIG. 3 is a schematic configuration diagram of an example in which the present invention is configured as a temperature or pressure sensor device,
As a conventional example, FIG. 4 is a structural diagram of a main part of a magnetic sensor device utilizing the Hall effect. In the figure, 10 is a semiconductor sensor device, 11 is a semiconductor substrate, 20 is a current path, 21 is a core of a conductivity type opposite to the current path for reducing the thickness of the current path, 31, 32 are signal extraction regions, 40 Is a fluctuation amount detection area.

Claims (1)

[Claims] 1. A current passage (20) inside the semiconductor substrate (11) extending in the thickness direction of the semiconductor substrate (11); the current passage (20) itself, or the current passage (20). 2
A variation amount detection region (40) provided in the middle of (0) for detecting a variation amount applied from the external environment; provided at both ends of the current passage (20) and biased to the variation amount detection region (40). Terminals for passing current (25,26)
And; electrically connected to the fluctuation amount detection region (40) inside the semiconductor substrate (11) and extending from the connection portion in a direction orthogonal to the current passage (20) to a signal detection terminal region. (33,34), and the above fluctuation amount detection area (4
0) from the fluctuation amount detection area (40) to the signal detection terminal area (3).
A semiconductor sensor device comprising: a signal extraction region (31, 32) for transmitting to a signal (3, 34).